Fix translation loss during catalog updates

README.md

@@ -233,10 +233,12 @@ FileNotFoundError: [Errno 2] No translation file found for domain: 'humanize'
 How to add new phrases to existing locale files:
 
 ```sh
-xgettext --from-code=UTF-8 -o humanize.pot -k'_' -k'N_' -k'P_:1c,2' -k'NS_:1,2' -k'_ngettext:1,2' -l python src/humanize/*.py  # extract new phrases
-msgmerge -U src/humanize/locale/ru_RU/LC_MESSAGES/humanize.po humanize.pot # add them to locale files
+scripts/update-translations.sh
 ```
 
+This extracts new phrases from the source code into `humanize.pot`, merges them into
+each locale's `.po` file and compiles the binary `.mo` catalogs.
+
 How to add a new locale:
 
 ```sh

scripts/update-translations.sh

@@ -1,7 +1,17 @@
 set -e
 
 # extract new phrases
-xgettext --from-code=UTF-8 -o humanize.pot -k'_' -k'N_' -k'P_:1c,2' -k'NS_:1,2' -k'_ngettext:1,2' -l python src/humanize/*.py
+xgettext \
+    --from-code=UTF-8 \
+    -o humanize.pot \
+    -k'_' \
+    -k'N_' \
+    -k'P_:1c,2' \
+    -k'PS_:1c,2' \
+    -k'NS_:1,2' \
+    -k'_ngettext:1,2' \
+    -l python \
+    src/humanize/*.py
 
 for d in src/humanize/locale/*/; do
     locale="$(basename $d)"

src/humanize/i18n.py

@@ -179,6 +179,19 @@ def num_name(n):
     return singular, plural
 
 
+def _pgettext_noop(msgctxt: str, message: str) -> tuple[str, str]:
+    """Mark a contextual translation without translating it.
+
+    Args:
+        msgctxt (str): Context of the translation.
+        message (str): Text to translate in the future.
+
+    Returns:
+        tuple: Original context and text, unchanged.
+    """
+    return msgctxt, message
+
+
 def thousands_separator() -> str:
     """Return the thousands separator for a locale, default to comma.
 

src/humanize/number.py

@@ -7,9 +7,11 @@
 import bisect
 
 from .i18n import _gettext as _
+from .i18n import _gettext_noop as N_
 from .i18n import _ngettext, decimal_separator, thousands_separator
 from .i18n import _ngettext_noop as NS_
 from .i18n import _pgettext as P_
+from .i18n import _pgettext_noop as PS_
 
 TYPE_CHECKING = False
 if TYPE_CHECKING:
@@ -35,18 +37,43 @@
 }
 _SUPERSCRIPT_TRANS = str.maketrans(_SUPERSCRIPT_MAP)
 
-_ORDINAL_SUFFIXES = ("th", "st", "nd", "rd", "th", "th", "th", "th", "th", "th")
+_ORDINAL_SUFFIXES = {
+    "male": (
+        PS_("0 (male)", "th"),
+        PS_("1 (male)", "st"),
+        PS_("2 (male)", "nd"),
+        PS_("3 (male)", "rd"),
+        PS_("4 (male)", "th"),
+        PS_("5 (male)", "th"),
+        PS_("6 (male)", "th"),
+        PS_("7 (male)", "th"),
+        PS_("8 (male)", "th"),
+        PS_("9 (male)", "th"),
+    ),
+    "female": (
+        PS_("0 (female)", "th"),
+        PS_("1 (female)", "st"),
+        PS_("2 (female)", "nd"),
+        PS_("3 (female)", "rd"),
+        PS_("4 (female)", "th"),
+        PS_("5 (female)", "th"),
+        PS_("6 (female)", "th"),
+        PS_("7 (female)", "th"),
+        PS_("8 (female)", "th"),
+        PS_("9 (female)", "th"),
+    ),
+}
 _APNUMBER_WORDS = (
-    "zero",
-    "one",
-    "two",
-    "three",
-    "four",
-    "five",
-    "six",
-    "seven",
-    "eight",
-    "nine",
+    N_("zero"),
+    N_("one"),
+    N_("two"),
+    N_("three"),
+    N_("four"),
+    N_("five"),
+    N_("six"),
+    N_("seven"),
+    N_("eight"),
+    N_("nine"),
 )
 
 
@@ -110,7 +137,7 @@ def ordinal(value: NumberOrString, gender: str = "male") -> str:
         return str(value)
     gender = "male" if gender == "male" else "female"
     digit = 0 if value % 100 in (11, 12, 13) else value % 10
-    return f"{value}{P_(f'{digit} ({gender})', _ORDINAL_SUFFIXES[digit])}"
+    return f"{value}{P_(*_ORDINAL_SUFFIXES[gender][digit])}"
 
 
 def intcomma(value: NumberOrString, ndigits: int | None = None) -> str:

tests/test_i18n.py

@@ -3,20 +3,72 @@
 from __future__ import annotations
 
 import datetime as dt
+import gettext
 import importlib
-import pathlib
+import shutil
+import subprocess
+from pathlib import Path
 
 import pytest
 from freezegun import freeze_time
 
 import humanize
 
-LOCALE_DIR = pathlib.Path(humanize.i18n.__file__).parent / "locale"
+LOCALE_DIR = Path(humanize.i18n.__file__).parent / "locale"
 
 with freeze_time("2020-02-02"):
     NOW = dt.datetime.now(tz=dt.timezone.utc)
 
 
+@pytest.mark.parametrize("locale, one", [("de_DE", "eins"), ("fr_FR", "un")])
+def test_update_translations(
+    tmp_path: Path, monkeypatch: pytest.MonkeyPatch, locale: str, one: str
+) -> None:
+    for command in ("bash", "xgettext", "msgmerge", "msgfmt"):
+        if shutil.which(command) is None:
+            pytest.skip(f"Translation updates require {command}")
+
+    root = Path(__file__).resolve().parents[1]
+    source = root / "src" / "humanize"
+    destination = tmp_path / "src" / "humanize"
+    messages = destination / "locale" / locale / "LC_MESSAGES"
+    messages.mkdir(parents=True)
+    for path in source.glob("*.py"):
+        shutil.copy2(path, destination)
+    catalog = messages / "humanize.po"
+    shutil.copy2(source / "locale" / locale / "LC_MESSAGES" / "humanize.po", catalog)
+    binary = catalog.with_suffix(".mo")
+    subprocess.run(["msgfmt", "--check", "-o", str(binary), str(catalog)], check=True)
+
+    results = []
+    try:
+        for updated in (False, True):
+            if updated:
+                subprocess.run(
+                    ["bash", str(root / "scripts" / "update-translations.sh")],
+                    cwd=tmp_path,
+                    check=True,
+                )
+            with binary.open("rb") as stream:
+                translation = gettext.GNUTranslations(stream)
+            # Replace the cached catalog so the second pass uses the updated one.
+            monkeypatch.setitem(humanize.i18n._TRANSLATIONS, locale, translation)
+            humanize.activate(locale)
+            results.append(
+                [humanize.apnumber(value) for value in range(10)]
+                + [
+                    humanize.ordinal(value, gender=gender)
+                    for gender in ("male", "female")
+                    for value in range(10)
+                ]
+            )
+    finally:
+        humanize.deactivate()
+
+    assert results[0][1] == one
+    assert results[1] == results[0]
+
+
 @freeze_time("2020-02-02")
 def test_i18n() -> None:
     three_seconds = NOW - dt.timedelta(seconds=3)